Nerd Talk: Can we predict team success?

Field Yates, Bleacher Report

Long, long ago, this league has seen our share of great players, teams coming out of nowhere, users over/under achieving, and even Beech and Pat Super Bowl wins. The montra has always been "any given Sunday" in that it can always be your year. And all too many times, user skills has trumped ratings in so many ways. I mean, let's look as recent as last cycle when Biggs was out here winning ships with the worst roster in football. Now many of you have seen my tidbits on the importance of QBs in team success, but is there more that we can attribute to outside of player skill (which is pretty hard to assign a value to)?

That being said, I ran a multiple regression model looking at not only the team's QB OVR rating, but also the team rating as well. I want to answer the age old question we always hear: DO RATINGS ACTUALLY MATTER?

I am here to tell you folks, I think they play a very large part.

**Now keep in mind, this season has been a huge tank fest and I truly believe some of these bottom teams could have won more games, which I would predict would only help the model even more, as some of these underperforming teams in red should be closer to their predictive model equation.

After entering the data in, the model results are as follows:

For a game that in theory should have a feeling of unpredictibility each week, I am not too upset to see an r^2 of 0.4507, which shows there are definately some large factors playing into our games. RATINGS AND TEAM BUILDING MIGHT ACTUALLY BE A NECESSITY! Now, I know not everyone in here may quite know how that model equation works, so let's take an example. The Denver Broncos have a 61.5% winning percentage currently. Their QB is 90 OVR and their Team OVR is 82. Let's compute how the model predicts their record:

Winning % = (-2.8746) + (0.0104 * 90) + (0.031 * 82)

Winning % = 60.3%, for a difference of just 1.2%!

Now the Broncos fit the mold very well, and not every team will fall right in line with the model, as there are still external factors of schedule/favorable divisions, player skill, tanking, playing teams that are tanking, etc. However, it is kind of fun to see that you can't, in a lot of cases, just pick up the sticks with anyone on the field and expect to win. Team building (strongly including the QB position) should be an important aspect to this cycles success, and I can't wait to see how some of the best team builders (and worst!) will fare moving forward.

Lastly, I know you all want to see where you all stand in the model and don't want to do the math yourself, so I have provided the chart below. PLEASE NOTE: The QB OVRs are from a couple of weeks ago, so dont @ me saying "My QB is an 82, not an 81!" You may interpret the [+/-] column as over/under achieving with your group, with positive numbers in green excelling above their expected results, but keep in mind some of those external factors listed earlier.

Enjoy!